SN74HC244: Unexplained increase in failure rate after adoption of RoHS-compliant materials and methods.

Thank you. I have contacted TI and been directed to start with the component supplier - unless there is a faster method to get an investigation started.

I'm glad you let me know - I can order samples and start looking at the performance of the parts much sooner than usual. For a full FA, we have to go through the usual channels (I'm just one link in the chain). There's typically a lot that happens before it gets to my desk.

I am sure we could also share the SAGE failure report if that is of interest to you.

Paul

I would recommend submitting that along with the FA so it is visible at all levels that check into this.

Will do. Contacting our supplier soon (Arrow).

Paul

For clarity, the exact TI part number we are dealing with is CD74HC244E.

A similar problem exists with SN74HC161N (this is also the exact TI number from the IC's).

We have contacted Arrow on the CD74HC244E.

Paul

Is it possible to get a definitive date-code decoding document?

Paul

I do have two questions. It seems that if I look up the latest material information on the two chips in question--74HC161 and 74HC244--they are currently listed as having gold bonding wires. I have also read much about TI's move to copper since 2008. Our report from SAGE indicates 2015 parts are copper and 2016 are gold - all manufactured in Malaysia.

Is there a trend here? Is there a move back to gold?

Paul Klos

Hey Paul,
I would have to check with my manufacturing team to find out why that change happened - I can't even confirm that it did happen without digging quite a bit. This is definitely something the QA team will want to look into though.

We have continued with our third-party analysis (as your distributor has not responded to our request for a FA as of yet) on these parts. The newest images demonstrate missing material from the copper bonding wire at the wedge bond. Multiple bonding wires appear to show material loss due to corrosion (or other phenomenon). As the design is mature (25 years-old), and previous failure rates were infinitesimal in nature, we are leaning towards a component problem. We cannot rule out RoHS processes--as they are more recent adoptions--but are left with our evaluations at this time. A high-pressure push with the distributor will occur tomorrow in an attempt to overcome their reluctance to participate (they have stated that the TI Community is our only forum to address the failure).

Paul

Thanks Paul,
I'm sorry to hear that the process isn't going smoothly. I will directly tie in our quality team to see what they can do to help. Can you tell me which distributor/office you are in communication with?

Arrow Electronics 
26632 Towne Centre Drive, Suite 100 
Foothill Ranch, CA 92610 

 949-375-4490

Manny

Hey Paul,
Can you also tell me what company you are with?

Hello Emrys,

I am an independent contractor working for Pacific Power Source in Irvine, CA.

I just met with our Arrow rep and provided initial data (our three third-party FA's and some initial application info like in situ pics, an assembly drawing, and a schematic).

Paul

Thanks Paul,
I will see what we can do to help from our end.

The parts finally went out for the FA. It took a while for Arrow Electronics to get that all organized.

Back on the subject of copper v. gold bonding wires, has there been a trend to return to gold?

Paul

Hey Paul,
I'm glad the FA has been started.

I can't really speak for TI on manufacturing trends, but in my opinion it makes sense to use copper where it works well, and gold where it doesn't.

Hello,

Is it possible to request date code information that would indicate whether an IC is built with copper or gold bond wires? For example, for CD74HC244EE4?

Sincerely,

Paul Klos

Hey Paul,
I looked into it and found that we don't keep track of which devices are built with copper or gold on our end (for the CD74HC244EE4). This method allows for flexibility in sourcing at our manufacturing sites, but could get confusing if you're trying to tell one from the other (like what you're doing).

If we have the full topside marking (date code and lot trace code) then we can look up the part and determine how it was manufactured/exactly what was used.

Hello Emrys,

Thanks for that Emrys. We will likely avail you of that assistance.

What I can say is that I have figured out, to my satisfaction, how to tell where and when a particular device is manufactured. If you are telling me now that the lot trace code provides further manufacturing data, I would ask if there is a means by which we could get decoding information for that as well.

Sincerely,

Paul Klos

Hey Paul,

There are a lot of details in this application report: 

It's actually from National Semiconductor, but it seems that we've kept the same markings alive after acquiring them.  Is there a specific LTC you're trying to decode?

FYI - each site can manufacture lots with different materials - as long as they are within the allowed bill of materials from TI. So, even if you knew which part of the LTC meant, for example, a part came from a Dallas FAB, there's still no way to be sure if it was made with copper or gold bond wires without us checking into it (or if you tore it apart and checked one from each LTC).

Hello Emrys,

Thanks for that information.

With regard to a specific LTC, we would like to understand the LTC's significance and specify the data and lot codes rather than interpret them after material receipt. In other words, we are looking for a method of specifying specific construction details as part of our material purchases.

I suppose I can do like I have today--for HC and HCT logic IC's we use--and look at the material references on the TI website for each part number we specify. Right now I am seeing a lot of gold for Malaysia-fabricated parts, and copper and gold in Aguascalientes. I might elect to purchase by part number and data code, where the specific date code portion identifying the manufacturing country is identify.

Sincerely,

Paul Klos

Sorry for the spelling errors above as I rushed the email.

The questions that I am trying to pose are these:

If we purchase by part number and date code, can we use the TI website material content data as a guide to get the internal construction materials desired?

Can we get a universal LTC decoder to at least tell us what bond wire material was used?

Sincerely,

Paul Klos

Hey Paul,
Please send that straight to Danytza at the email I provided. I would just ask her anyways, and this will speed things up for you.

My apologies; I missed that email reference. Will do.

Paul

We are a small contract manufacturer and I was making a quick search for open issues with 74HC244 - our particular usage is with the SN74HC244DW. It seems to be either a output drive current change or a timing alteration (slew rate or propagation time). It's still early in investigation, but it crosses at least two mfg date codes. We've not had a problem with the part before and this is a mature product that it is going into. I'll look into the mfg code on the ICs to try to narrow our focus, but it is interesting to find another instance of massive failure increase based on a new lot code. My initial question would be if there was a new manufacturer location / process using a new die size. We've experienced issues from smaller die sizes in the past with input slew rate changes.
Any feedback would be appreciated as we work with our vendor (I think it is Arrow) about investigating the cause for failure.
-Mark

Perhaps the TI employees/bloggers can be of more use to you. All I can tell you is that we had 2015 through-hole parts that began failing. Further analysis led to our technician finding intermittent behavior at input and outputs. One moment they were conducting, and the next they were not. The difference turned out to be pressure imposed on external IC components leads by test leads - something we surmised due to laboratory analysis showing corroded stitch bonds where microscopic gaps existed in bond wire material. If the technician pushed hard enough, he may have been causing the gap to close and create partial operation. Tested in-situ in their target PCBA's, the IC's appeared to have open inputs/outputs. The bind wires were copper.

Paul

Hey Mark,
I would recommend opening a new thread with details of the issues you are having. I will need to see exactly how the part is being used (schematic preferably) and how it's failing (scope shots preferably) to make any type of helpful determination.

That's an interesting thing! I'd always assume fractured blind via in the PCB from delam if pressing like that - I had a similar thing with a very heavily populated board with 16 layers - it would cost as much to have analyzed as it would rebuild. But I digress, this issue with the timing/drive isn't bond wires. I had a false positive on the first check - replaced a suspect part with a known good part and had good results. The second check didn't pass, so the fault is an adjacent device. We had run two MFG dates on one lot and 100% of our failures all contained the 2nd MFG date code, but apparently there is another factor. It's always a scramble to find a root cause when production dates slide right.
Thanks for the insight to the through hole device defects.